CIMI MTF Minutes 20130606

From CIMI
Jump to: navigation, search

CIMI Modeling Taskforce - Meeting Minutes

Thursday 6th June 2013 @ 20:00-22:00 UTC


Attending

Linda Bird

Harold Solbrig

Stan Huff

Rahil Qamar Siddiqui

Mike Lincoln

Patrick Langford

Joey Coyle

Michael van der Zel

Sarah Ryan

Dave Carlson

Patrick Langford

Mark Shafarman

Eithne Keelaghan

Agenda

  • Unit of Measure modelling style (continuation)
  • Implementation & Modelling Activity Updates:
  1. CIMI Reference Model - DSTU Publication on https://github.com/opencimi/ (EA, BMM, XMI, RDF)
  2. Archetype Modelling Language (UML)
    1. http://informatics.mayo.edu/CIMI/index.php/AML_Meeting_June_2013
    2. Next Thursday - we'll ask Dave to do a demo of the AML work
  3. IHTSDO workbench
  4. Mindmap-to-ADL generation
  5. Kaiser UML-modeling exercise
  6. Other tooling updates?
  • Brief Github tutorial (Harold)
  • CIMI URIs (Harold)
  • CIMI face-to-face agenda
  • CIMI Model Example Instances (generic, model-specific, green)

Unit of Measure (modelling style and approach)

  1. Finalise discussion from last week:
      1. CIMI general models will define specific unit of measure where a global standard is in place (e.g. Systolic blood pressure in "mm[Hg]")
  2. Mindmaps to ADL conversion (and instance generation?)
      1. Where a global standard for units of measure is not in place (e.g. Body temperature), CIMI general models will only bind data elements to the Property of each quantity (e.g. mass concentration, time, volume, length, pressure) and a reference set of valid alternative units for that property (without defining the specific unit of measure);
      2. Regional specialisations can specialise the CIMI general models to define specific units of measure used locally
      3. A 'CIMI Canonical model' will be defined, which specialises the 'CIMI general models' to the specific unit of measure to be used when interoperability is required between jurisdictions
        1. Canonical model may both restrict the UOM and define any appropriate min/max value constraints applicable to that unit of measure
    1. Units of measure will be included in Property-based reference sets (e.g. Temporal_UOM_refset)
  3. Finish discussion on Daniel's last comment last week, regarding it being difficult to enumerate all local units in the CIMI models (to ensure that the property-based UOM reference sets being complete) - e.g. Uncommon units, such as Measurement of Amount of resources. Suggestion: Standardise normal cases as much as possible, while keeping border cases open.

Peter Hendler/Kaiser exercise:

    1. Progress and plan for Diabetes UML modelling exercise
    2. Produce Kaiser UML version
    3. Produce AML UML version
    4. Compare overall tree structure

Kaiser defining their model in a spreadsheet -> mindmap -> AML tool


Detailed Meeting Minutes

Linda: Draft agenda - to finish unit of measurement. Might put off to wait for Daniel. Implementation and modeling activity updates. Harold will give Github tutorial. Virginia wants to know what we want to talk about at face to face meeting. CIMI model examples instances - Michael has (?) to show us. Are all OK with the agenda?

[no response]


Kaiser UML-modelling exercise - Stan Huff

Linda: Any updates on the Kaiser exercise, Stan?

Stan: Yes. Peter says they are working on it. I thought [just needed to] export representation of model from Protege, but they thought this wasn't reasonable. They will represent it in a spreadsheet and send it out. Not a high priority for them, so it will take a week or so.

Linda: So when we get the spreadsheet, we will do a Mindmap and send to folks so Dave can put into a tool or...?

Stan: We are looking to get AML representation of this - however easiest to get these. We can ask Dave.

Linda: Sure. Probably helpful to Dave to have modeling work to plug in. Can plan and talk to Dave... whether will work for him. Thanks, Stan.


CIMI Reference Model - DSTU Publication (EA, BMM, XMI, RDF) - Michael van der Zel

Linda (cont'd): So - CIMI Reference Model. Michael - you put a new version up?

Michael: Yes - 1.0.4 - 1.0.0-DSTU.4

Linda: So we have EA version?

Michael: Yes - I tried to generate... RDF...XMI...UML... The BMM - has some issues still. The XSD(?) - not using a namespace, but we could.

Linda: What issues are still outstanding on BMM?

Michael: I treated Boolean as String and another minor thing. There are 3 issues - Boolean, cardinality on single attribute, and parameter is wrong(?)...

Linda: OK - Harold - Github?

Harold: Thanks, Michael. Really appreciate that. Tell us when BMM and XMI(?) are up... when tools that consume BMM are...

Linda: OK - I might use this as an exercise to walk through... Github tutorial.

Harold: Yes.

Linda: Thanks, Michael.


Archetype Modelling Language (UML)

Linda(cont'd): Can anyone give AML update?

Harold: Dave not here. We put together (?) presentation for Tuesday in Berlin... Trying to optimize the time there. We put it out on CIMI Wiki, but have not linked to it. I'll put the [URL] on Chat.

http://informatics.mayo.edu/CIMI/index.php/AML_Meeting_June_2013

Linda: Thanks, Harold.

Harold: A week from next Tuesday... In concert with OMG meeting, Thomas Beale will be there... So time to get this together... Still working on it...

Linda: When will it be good for Dave to do another demo for the group?

Harold: Next Thursday. We are going to do a presentation to AHRQ people next Wednesday - right Sarah?

Sarah: Yes.

Harold: So we will put together reasonable approach to what we are doing and why.

Linda: Great! All will be interested in that.

Michael: I was entering the BMM issues as Git issues... Did not know you were getting... (?)

Linda: OK. Mike Lincoln?


IHTSDO workbench - Mike Lincoln

[Summary of text below: A contracting task order was submitted to VA to support the CIMI Workbench software.  This was placed in April and has a ca. 3-4 month turn-around time.  In the meantime there are publicly available VA training materials that could be used to train CIMI participants.  Mike Lincoln will keep the CIMI task force informed on the progress of the task order.]

Mike Lincoln: in April - I put in contracting (?) - the necessary... documents for workbench. These are task orders. They are placed against a contract - write individual task orders... buckets run into... then turn off.

Mike (cont'd): Money already obligated. But must wait for task order for level of approval - not high, but takes 3-4 months. I put in late April... so sometime in July or maybe August, we will find out. I will call contracting office and ask... They are limited to tell us... But they have not asked significant questions... We addressed those questions they asked... So I am hoping it will be approved.

Mike (cont'd): Harold asked if we can start before approval. No - since we can't do non-contract work. But I will talk to Catherine Hoang and see if there is an existing contract. But there is some training we can do. Harold already participated in that. Questions?

Harold: To the group: What do we need to be doing in the interim? Are there needs we have before then... and can we get them listed... stop gap before we get these in place?

Mike: Yes - other needs... All I can think of is training.

Harold: Last week - was concerned about getting started. Depends on when we will start putting together value sets, but maybe not till July.

Linda: Yes, and stop-gap is Excel.

Harold: How much have we used Excel?

Linda: A little but not a lot. What Rahil and I put together.

Harold: OK.

Linda: Any other updates on tools... or implementation?

[no response]


Mindmaps to ADL conversion (and instance generation?) - Patrick Langford

Linda: Patrick - Mindmap update?

Patrick: I had done some work with the original 16 Mindmaps, and I thought that was most of them. So it was a bit of a shock when I saw 60+ Mindmaps on the Google drive. So, Linda, you gave these to me and the syntax looks great. I've rewritten my parser to work with this syntax. One problem - since I thought this would be a one-time conversion and go past Mindmaps, I didn't build the concept of repositories the way the ADL Workbench uses them... with 60+ Mindmaps, and since there will probably be more, I need to re-think how the converter will work.

Linda: Good.

Patrick: Yes - only a few to inconsistencies and will be easy to fix. Harold gave me Github info... I did not put up files. Want to rename. Currently CIMI-classname"dot" ...

Patrick (cont'd): I want to use "-RM-". ADL workbench requires this. When I have the Mindmap, the file name is one of the only places I can get this information... Would that be OK? One of the canonical places for these and should be reasonable...

Linda: Yes. I want to understand. Give an example.

[slide] - Mindmaps: CBC Lab Test Observation

Linda: So the package would be CORE for us followed by entry...

Patrick: I didn't try CORE. What is required is RM.

Linda: ?

Patrick: Since I am re-writing my tool... I already have package name in BMM. I will change the file back and put them in the git repo... So they will be up there shortly. OK?

Linda: Sure. Should ask the group. The archetype standardization language reference model followed by CORE followed by entry.

Patrick: That is in the BMM. Are you saying we should change the name of the file?

Linda: Yes - now the... If we are sticking to standard... Add the CIMI-CORE entry... Distinguish package... from Cluster... Any comments whether we should... Stick to archetype - identification package and add CORE...

Linda (cont'd): Michael - are you happy with package entries and Cluster referred to a "CORE"? Now we have parties and packages?

Michael: Now I call them "CORE" and "PARTY" and... RM. Should I rename?

Linda: As long as we are consistent.

Patrick: Yes - have to have the right BMM and referenced files and Ref-ids. They have to match - whether CORE or RM... We have to decide.

Linda: I would suggest "CORE" and "PARTY".

Michael: So I have to rename.

Linda: OK - all OK? So, I rename to... CIMI-CORE-entry... and where we have... demographics... CIMI-Cluster... to distinguish PARTY clusters from CORE clusters. All happy?

Answer: Yes.

Patrick: Are you going to rename inside the files as well?

Linda: Oh... Is there a fast way to do that?

Patrick: Yes - don't change. I will do this... and push up to Git-repository(?)...

Linda: Thank you so much, Patrick. It would be great if we could do in parallel to this. If you can make changes...

Patrick: I will have up before meeting ends.

Linda: Thank you, Patrick. Anything you need?

Patrick: Yes - looking at options to go forward. One thing for the Mindmap to ADL option... I've been looking at FreePlane since it supports plugins... Might make the Mindmaps more usable... Might have a solution that would help make these more manageable...

Linda: I would love plugins. Thank you, Patrick. Questions?

Rahil: I wanted to know... Github... Are we all on Github now?

Lind: Next is a tutorial on Github by Harold. Would that work for you?

Rahil: Yes.


Brief Github tutorial (Harold)

[slides] - Harold

Harold: First you figure out what is going on - easier. Version control involves notion of central repository. Then check things out... Some pull out locally and you... create edits... Then commit back...

[slides]

Harold: Repository maintains... If read only - periodically do an update. In Github - it is federated with repositories. It is powerful. You can establish version control with little effort. You type "git init" and...

Harold (cont'd): It is a local repository. You can edit... find out changes... You can deal with it as if it was a local repository. Another thing - if you have the URI and you have permission to read, you can do a Git-clone and it creates a central repository(?). It contains "this is where I come from". And you can make changes... it is your own... "git clone".

Harold (cont'd): The first thing is - where do I get some clients? Git is easy to use and easy to find on Google. Set of GUI's... I don't use the GUI's... it is worth it to understand what is going on underneath, so you understand...

Harold (cont'd): CIMI is in Github - a collection of repositories. When Thomas Beale was putting together, I ended up as czar... and (?) is co-czar... and Michael van der Zel is (?)czar.

Harold (cont'd): If you want to clone - it is there. You do not have to do anything to get out of this. I clicked on (?) earlier... Interesting stuff... If I want to clone it - "Git Read only" gives read-only. If on a Mac, you can punch this button... Github gives you other interesting things. "Pull request", "Issue-log", "Wiki"...

[slide]

Harold (cont'd): Git Basics - some things you can do. Clone - makes a copy of an existing repository. Will include all the files and a way to manage locally. "Status" - what has changed in my copy. "Add" - add a file to local repository. "Commit" - changes to local repository. "RM, MV" - can remove a file.

Harold (cont'd): Git carries a history of everything that has changed. "Log" - commit history. "Remote" - tells you where you cloned something from. These allow you to put repository together. "Pull" - fetch changes from remote and merge. If collisions, up to you to fix. If you pull it keeps you current.

Harold (cont'd): "Push" - take the work I've done and copy to remote. "Diff, Log"...

Joey(?): My confusion with "Add" is- add a change? If file is added and you change, you are adding the change. Your definition is a little misleading.

Harold: OK - I have an example.

Patrick: I think "Add the change" rather than the file.

Linda: Add a version of the file.

[Harold shows screen]

Harold: Well - So here is reference model. And it says... all these files are happy with local repository. So I have the file in the local directory, and it says "You have file in local directory". If I wanted to add, I say "Git...Add". So it is in, but it is not committed. So I say "Git...(?)". So now it is in the repository. So - it tells Git that this is something I want tracked.

Patrick: ...(?)

Harold: I don't think it will ever let you add change... Oh - I see. You are right.

Patrick: You are not adding file to repository... You are saying there is a change... You are adding "hyphen A" . Says "Add all things that have changed"... It is dangerous. If you have changed many, than it adds all... and you may not want that.

Harold: Oh... A nuance I was not aware of.

Linda: This is from Github shell?

Harold: No. On windows client... GUI clients can make life easier... Gather info and...

Michael: I agree. Windows client... I removed... Eclipse... Now... I thought Dave was preparing...

Harold: Thanks for update on Add. Let me take us through operations. Then I can turn over to someone with windows client. When I do status...

[screen - demo]

Harold (cont'd): ...I can push this change out here. I can push out and undo. A safe environment... Easy to fix if there is damage.

Harold(cont'd): Cloning a (?). Git clone... So - if want read/write access...

[Harold continues with demo]

Harold: At any point I can do "get pull" if I am concerned whether is current. All sorts of options on Log.

[Harold continues with demo]

Harold (cont'd): So I went through "Clone the repository". The first thing I did... Let's say I got rid of... Remove file... Let's say I did not want to remove... I can... and got it back. If I want to remove... Git remove... It will remove not only in your directory, but in repository. So I did... and it said... Commit... git commit-m "added a silly file"...

Harold (cont'd): Want to put onto our Github. You can't push unless already up-to-date with where you are in repository. So here - Michael submitted - this afternoon... If collision - there are tools to get edits combined. You are best off coordinating.

Comment: Yes - is binary. So is one big blob.

Harold: Yes... So I did "git pull", then "git push". Then I can see "silly file" in Repository. So I did "remove"... So, instead of removing, I edited... added.

Harold (cont'd): The key is - when you do an init, you have local repository, and that is what you work with... You make changes and when done, you can push out to central repository. Right now, we have 2 directories. 1.0.3 and 1.0.4. You can tag a particular version with a <tag>. Can list the tags...

Harold (cont'd): git tag (signing option) - m "message" <tag>... Can tag later (see ref). Share by "git push origin <tag>". Many projects use signed tags as a way of stamping "Officially blessed releases".

[slide] - has the following:

In summary:

Everyone has their own repository

Git is great for managing your own workspace

Git clone is a clone with a link to where you got it


[slide] - has the following:

"Server" is a matter of community agreement.

-github is often a good choice. Can tag, branch, merge and do all sorts of useful stuff.


Harold: I haven't answered Linda's question.

Linda: I would like to download a client and clone a directory and make changes and push them back.

Harold: I don't have a Window's client.

Linda: I do. Michael can explain why he didn't like it.

Patrick: Name of client?

Linda: I will show you.

Patrick: There is... I can search for Window's client. Don't want to use command line - will cause you grief.

Linda: Michael?

Michael: ...?

Patrick: A plugin?

Answer: Yes.

Michael: The windows client is a Git client. Not like it... Others may... So Eclipse...

Patrick: Most will be like that. Revolves around making Git easier to use. The Eclipse plugin is integrated. One thing - certain instances when it can be a bit of a problem. Harold - we can talk about it.

Does CIMI have a License for what is put out on Github?

Harold: To get Push permission, you need to be added. If not have this, you don't have to worry. So write to (?) or (?) or me and we will give you permission. So if willing to share, then is free. If pay a nominal fee - then can be private.

Michael: It is public, but we don't have any license.

Harold: Need to do this. Didn't we settle on one?

Michael: I think so, but I didn't put in... So not feel good about this.

Linda: Would be good to know what I can do... in standard...

Michael: I think I... [can't hear] I don't know enough about copyright... So not feel comfortable...

Harold: Didn't we do that in one of the meetings?

Linda: Not when I was there.

Patrick: If other stuff people have done... I was not around for discussion.

Harold: If hasn't been done, we will need to do it.

Sarah: I will help too.

Linda: Thanks, Sarah.

Mike Lincoln: These are public. In your searching for CIMI, will see these now?

Harold: Yes.

Linda: I have downloaded these... But out of date... How do I get my local clone in (sync?)... Local version?

Michael: One of the things. I could not get my head around this...

Patrick: See the "sync"... at top - click on this.

Linda: OK. Is syncing pushing my changes?

Patrick: Did you make any changes?

Linda: I hope not. Oh - this is the new version. I see Dave Carlson on the call.

Stan: And I just came back.

Linda: And Dave - you were volunteered.

Dave Carlson: I will be on a plane.

Linda: We were hoping what you will give in Berlin will be...

Dave: June 27th.

Linda: 3 meetings away.

Dave: When I will next be available.

Rahil: That is the CIMI face-to-face meeting.

Linda: So next Thursday - you are away... and then on a plane?

Dave: It is possible... in Chicago... But I don't know if I will have an Internet connection.


Dave Carlson - Presentation next Tuesday, June 11th

Stan: What about next Tuesday?

Linda: Yes.

Harold: We have been talking about update on (?) on Tuesday.

Linda: Do we have time for both?

Harold: Let's talk about this. Dave - can you do this?

Dave: I need to prepare for this anyways for Berlin. So I will plan on that.

Linda: Great. Thank you, Dave.

Linda: So here is DSTU-4 version... Thanks for your work on this Michael. So - still going to do UID's derivable? Not sure if we need...

Michael: I am not sure either.


Decision about Phases and Versions of DSTU

Linda: I would remove it... optional... Thanks for keeping Reference Model in sync. Should we be publishing a single DSTU so people know if DSTU 4 or DSTU 5? Or should we (?) and then inclement the version?

Dave: I would suggest using the (?). Until we get to (?). I will not look at it.

Linda: How long do we need to be in DSTU-phase before we go on to (?)-phase?

Dave: Analogous to Beta?

Linda: I think version 2, 3, 4 - people will think is changing, like you say.

Michael: We should make sure the documentation is correct.

Patrick: DSTU is more of a Beta... Should not be release candidate until used...

Linda: That is what we were thinking... Baseline DSTU... Need to have a stake in the ground. Opinion on what we should call the DSTU? Dash-DSTU?

Michael: Yes. Remove all different versions... Minor... Release candidate - create a new folder.

Linda: That would be my preference... Once released, need to track all the small changes... Published changes. So all happy? Dash-DSTU... Baseline stake in... Use in AML-tooling and other tooling. All OK?

[no response]

Linda: OK.


Update on AML - Dave Carlson

Dave: Demo-specification... Brief update. Harold and I... started a Wiki-page with questions and notes... Face-to-face in Berlin... will go out to the group... One thing - multiple destinations for any model-element... Can have both a name and description... Can have translations into multiple languages. No (?) in AML to support this (?).

Dave (cont'd): One thing we are working on is ISO 11709...3 - a good meta model... recent... designation. Designable... Tested yesterday. Idea of archetype stereotype. Archetype inherits from 11179... Can have multiple designated elements... Working on... critical... Not have ability to translate names... Going to work well... Also did not have ability to have language designations... I can demonstrate on Tuesday... CTS2... Now can be imported to UML... Implementation of 11179. I think it will work out well. Standard for designations. Profile implementation for that.

Linda: Great. I particularly liked transform between AML and UML...

Dave: We had... OMG... Pull out into UML... [did not hear well]

Linda: Thanks for update. We look forward to Tuesday's presentation if you are available.


Planning for Face-to-Face Meeting in 3 Weeks

Linda (cont'd): Only a little time left now. Hoped we could brainstorm for the face-to-face meeting. We have 3 weeks until that meeting. It would be great to be able to display some of the Mindmaps. What is possible? Have we the AML work to present, Dave?

Dave: Yes - it will be remote, but I will plan to do...

Patrick: I had 16(?) converted. They are still up on the URL... I am now building a tree view... XML... Bring in some of icons from workbench. Would those be...?

Linda: Ideally - Take the ADL generated and...

Patrick: Yes - those will all come up and you can view in ADL workbench.

Linda: Michael - How long until sort out problems of BMM?

Michael: Just have to find time since it basically works.

Linda: 3 weeks?

Michael: Yes. I don't know if it is a problem with (?) and openEHR...

Linda: I think in 1 tool would be ideal.

Michael: Should ask Thomas or the Spanish guys... current BMM?

Linda: After changes?

Michael: Yes...

Linda: Can you follow-up with Tom and (?)?

Patrick: For current ADL, I will... and see if I have any problems.

Linda: Great. And when Michael gets... Patrick - you are testing in current...?

Patrick: Yes.

Linda: Can you send instructions to all?

Patrick: Yes. I gave to Intermountain and he could bring it up. Especially if running Windows... Should work.

Linda:...

Patrick:...

Linda: So - at the Face-to-Face, I will not be there. I don't think there is as much to present on modeling. Probably - a brief update and then implementation work we have been doing. Suggestions?

[no response]

Linda: And workshops?

[Linda adds to her notes]

      • Brief overview of activities
      • - Example instances

Linda: Stan - what other topics?

Stan: I am thinking of other stuff... modeling stuff...

Linda: Feels like is a short window between meetings. We have been focusing on... implementation activity rather than modeling...

Stan: A way to maintain terminology... So we can create CIMI reference sets and connect to models. Kind of vague... Harold may be able to say a better list of actions.

Harold: One of the things we need to settle on the details as far as URI's... coded representation... terminology binding... some is AML-work. Some are issues we are encountering with binding itself.

Linda: You talking about coded representation of coded text?

[see Linda's notes]

Harold: Yes. There will be stuff that is difficult to translate... RDF... And talk about some of the things we inherited from 21909-CD data types.

Linda: Yes... when populating the coded text... URI's for coded... Which policy we are following... HL7...

Harold: We can specify... world of RDF... Do we really want to do it these days? We can go over that whole set.

Stan: I love examples.

Linda: Examples Rock!"

Comment: There are 2 types of transforms. One is representation in... a green(?) stripped-out version. And... in XML and then have transforms... JAVA instances... is that what...?

Harold: Java... not...

Linda: XML and RDF...

Harold: Yes... Some are just transforms. The HL7 way but kind of a ... way of doing things. In most communities where EN is sufficient... it is... how much you have to put in.

Linda: Yes. In NHS, Nicholas was pushing for having everything in instance... Whereas Intermountain... realistic...

Harold: ...how to represent English language. I think we are erring on wrong side of model.

Michael: ...transformations between... Talking about Example... instance format.

Linda: Also - last week. I presented... That explains the language I was using. But you are right. The instance examples we should send around to the group.

Michael: OK, but I want some actual clinical examples. I did technical part... the XSD... if you can point me to the Mindmaps I can use...

Linda: Hopefully I can point you to them on Github... generating... example instance... templates for easy populating... Mock interfaces would be... but populating...

Linda (cont'd): Now have 5 more minutes. Any other topics for Face-to-Face? What about getting back to modeling activity? Receiving patterns... discussion about observation... Whether we will start on immunization... has taken a back seat.

Stan: I agree with that.


CIMI Website and Portavita - Sarah Ryan

Harold: What about website?

Sarah: I've been in touch with Portavita guys and there will be a presentation.

Harold: Do you want time Tuesday or wait till meeting?

Sarah: Maybe want some time on Tuesday... in case of problems.

Harold: So - give them a slot on Tuesday.

Sarah: So I will get in touch with... Amsterdam.

Linda: Thank you. Would be part of broader presentation... A part of this... What is the site going to look like? How to browse to model... Also - we said we would standardize some visual representation of model. HTML-style and... So, regardless of which model, need graphical ways to view model...

Linda (cont'd): You raise a good point. There should be a more clinician-friendly way to view models. Need to get to this... ASAP.

Stan: Yes.

Linda: I will send out draft list to everyone and if have anything to add, let me know. So - next Tuesday - interesting things.

[end-of-meeting]